Humans have been, are, and will forever be thirsty to invent things that would make their lives easier and better by a thousandfold. The capacity of what a human mind can do has always baffled me. One such major invention would be what is called as AI- Artificial Intelligence. Wouldn’t it be great if machines could think? That’s precisely what AI is. We humans have natural intelligence. But if machines can think, it’d be artificial. So, AI is just a collective term for machines that can think.
Now here are some examples of AI in real life. Robots are what come to mind first. They are machine replicas of human beings. They can think for themselves, take important decisions on their own without human help. Not all artificially intelligent machines need to look like human beings though. Some of the other examples include self-driving cars or Amazon Alexa or even Siri. One other important application can be speech recognition. Remember how you ask google by speaking instead of typing what you have to search for, into the search bar? That’s one of the applications right there. There are so many more applications but let me get on to other topics.
Alan Turing, a brilliant mathematician, who broke the Nazi encryption machine Enigma, came up with a history-changing question, “Can machines think?” in 1950. The actual research began in 1956, at a conference held at Dartmouth College (a lot of the inventions have come into the picture, thanks to the Ivy League). A couple of attendees at the conference were the ones who came up with the idea and also the name “Artificial Intelligence”. But since the whole idea was new, people didn’t buy the idea and funding for further research was pulled off. This period, the 1950s – 1980s was called “AI Winter”. In the early 1980s however, the Japanese government saw a future in AI and started funding the field again. As this was interconnected to the electronics and computer science fields, there was a sudden spike in those as well. The first AI machine was introduced to the world in 1997; IBM’s Deep Blue became the first computer to beat a chess champion when it defeated Russian grandmaster Garry Kasparov. And that, my dear readers, was the advent of a massive field called “AI”.
Basic Of AI
The basic function of the algorithms of AI is data analysis. Let me put it this way. How do you think human beings learn new things? They observe. They observe and that’s how they learn. Machines learn the same way. A huge amount of data is fed into the machines, and they observe and learn, observe and learn, observe, and learn. Since they are machines and don’t usually get tired, unlike humans, this process of learning is never-ending. Data that is fed into the machines could be real-life incidents. How people interact, how people behave, how people react etc. So, in other words, machines learn to think like humans, by observing and learning from humans. That’s precisely what is called Machine Learning which is a subfield of AI.
Types Of AI
AI can be broadly classified into two:
- Narrow AI: This type of AI is also referred to as “weak AI”. Narrow AI usually carries out one particular task with extremely high efficiency which mimics human intelligence. An example would be any computer game where one player is the user and the other player is the computer. What usually happens is, the machine is fed with all the rules and regulations of the game and the possible outcomes of the game manually. In turn, this machine applies these data to beat whoever is playing against it. A single particular task carried out to mimic human intelligence.
- Strong AI: Also referred to as “general AI”. Here is where there is no difference between a machine and a human being. This is the kind of AI we see in the movies, the robots. A close example (not the perfect example) would be the world’s first citizen robot, Sophia. She was introduced to the world on October 11, 2017. Sophia talks like she has emotions.
There are 4 distinct categories of AI namely:
- Reactive machines: These are the most basic type of AI and are purely reactive as the name suggests. They neither can form memories nor can use past experiences to form decisions. An example would be IBM’s Deep Blue chess-playing supercomputer which is mentioned above. Deep Blue beat the international grandmaster Garry Kasparov in 1997. It can choose the most optimal of the chess moves and beat the opponent. Apart from a rarely used chess-specific rule against repeating the same move three times, Deep Blue ignores everything before the present moment, thus not storing any memories. This type of AI just perceives the world, the chess game in the case of Deep Blue, and acts on it.
- Limited memory: These machines can look into the past. Not the ability to predict what happened in the past, but the usage of memories to form decisions. A common example could include self-driving cars. For example, they observe other cars’ speed and directions and act accordingly. This requires monitoring of how a car is driven for a specific amount of time. Just like how humans observe and learn the specifics. These pieces of information are not stored in the library of experiences of the machines, unlike humans. We humans automatically save everything in the library of our experiences and can learn from it, but limited memory machines can’t.
- Theory of mind: These are types of machines that can understand that people have beliefs, emotions, expectations, etc. and have some of their own. A “theory of mind” machine can think emotionally and can respond with emotions. Even though there are close examples of this kind of AI like Sophia, the research is not complete yet. In other words, these machines have a notion of not just the world, but also the existing entities of the world, like human beings, animals, etc. These machines will be capable of answering simple “what if” questions. They’ll have a sense of empathy.
- Self-Awareness: These types of machines can be called human equivalents. Of course, no such machines exist and the invention of them would be a milestone in the field of AI. These basically will have a sense of consciousness of who they are. The sense of “I” or “me”. Here’s a basic example of the difference between “theory of mind” and “self-awareness” AI. The feeling of I want to play is different from the feeling of I know I want to play. In the latter, if you notice, there is a sense of consciousness and is a characteristic of a self-aware machine, while the former feeling is a characteristic of a theory-of-mind machine. Self-aware machines will have the ability to predict others’ feelings. Let’s hope the invention is not so far away!
Today, the amount of data in the world is so humongous that humans fall short of absorbing, interpreting, and making decisions of the entire data, no, even part of the data. This complex decision-making requires beings that have higher cognitive skills than human beings. This is why we’re trying to build machines better than us, in other words, AI. Another major characteristic that AI machines possess but we don’t is repetitive learning. Humans are observed to find repetitive tasks highly boring. Accuracy is another factor in which we humans lack. Machines have extremely high accuracy in the tasks that they perform. Machines can also take risks instead of human beings.
AI is used in various fields like:
- Health Care
Disadvantges OF Existing AI Machines
- High costs of creation
- No human replication due to lack of emotions
- Zero creativity
- No improvement with experience
The Threats AI Poses
Even though there are various advantages to AI, they can someday maybe overpower human beings. That can be highly dangerous. Here are some risks or threats that AI poses in the future:
- AI can do something devastating: There are various applications of AI which are even used to design autonomous weapons and missiles. In the wrong hands, this could be highly devastating. Wrong use of AI could lead to an AI war too. This is not a present threat, however, because narrow AI is harmless. But this could be an increasing concern as the levels of AI increase.
- AI is programmed to do something but it develops a destructive method to achieve the goal: What we have in mind, can be extremely difficult to feed into the machines. Just like GIGO (garbage in garbage out), we need to be highly careful in aligning the AI’s goals to ours. For example, if you ask a self-driving car to take you to the airport as fast as possible, it might exceed the speed limit, make you nauseous due to the high speed, and can even land you in legal disputes due to the breach of the speed limit. Another example in the higher levels of AI can be – if you design an AI and ask it to take measures to balance the ecosystem, it might just go and kill some of the people to reduce the population to normal so that the ecosystem is balanced.
- AI can someday overpower humans: The reason why humans sit at the top of the ladder of all creatures is that we are the smartest of the species there ever exist. If we develop an AI which is smarter than us, it may pose a threat to humans. Various movies are based on this concept. Also famous scientists like Stephen Hawking, Elon Musk, etc are highly concerned about the same issue.
Let’s see how far we humans can push ourselves in creating art which doesn’t destroy us in the end. We are smart enough. This article is not to scare you of AI, but just to educate you. It’s important to know about everything in a field; pros, cons, threats, everything.